Choose the right mix of Claude Team, Claude Enterprise, Claude Code and the Claude API. Roll it out with SSO and connector controls in place. Connect Claude to SAP, NetSuite, Salesforce, HubSpot and your own systems through custom MCP connectors.
US-based leadership. International delivery. Your team, enabled.
“Get me ready for the next customer conversation.”
Illustrative workflow. Connections and actions are designed around your permissions and project scope.
Works with the business you’ve already built.
CHOOSE THE RIGHT STARTING POINT
Anthropic sells Claude as seat-based plans, as a developer platform and through cloud providers. Most organizations end up using more than one. This summary follows Anthropic’s documentation as of September 2026.
| Option | Best for | What to know |
|---|---|---|
| Claude Team | Teams of 2 to 150 that want chat, Cowork and Claude Code on one subscription. | SSO, JIT provisioning, central billing and admin controls for remote and local connectors. Standard and Premium seats can be mixed. |
| Claude Enterprise | Organizations with security review, audit and data retention requirements. | Adds SCIM, role-based permissions, audit logs, the Compliance API, custom retention and IP allowlisting. The seat fee covers access; usage is billed at API rates. |
| Claude Cowork | Handing Claude a multi-step task across files and connected tools, then reviewing the result. | Included on paid plans. Sessions run in Anthropic’s cloud by default. Anthropic is merging Cowork and chat into one app; Team and Enterprise keep both as they are for now. |
| Claude Code | Engineering teams working in the terminal, an IDE, the desktop app or the browser. | Included with every Team seat and with the Enterprise seat on new plans. Can also run through the Claude Console, Amazon Bedrock, Google Cloud or Microsoft Foundry. |
| Claude API (Claude Platform) | Building Claude into your own applications, portals and agents. | Billed separately from Claude subscriptions. Offers the Messages API, Claude Managed Agents and an MCP connector (beta) for remote MCP servers. |
| Claude through your cloud | Organizations with existing cloud commitments that want cloud billing and IAM. | Available through Amazon Bedrock, Claude Platform on AWS, Google Cloud’s Agent Platform (Vertex AI) and Microsoft Foundry. Features vary by platform. |
Plans and features change often. We confirm current terms against Anthropic’s documentation with you before scoping. Subscriptions and usage are billed by Anthropic or your cloud provider, separately from our services.
FROM POSSIBILITY TO PRACTICE
Start with a rollout, a connection, or a workshop. Bring us in where you need us. We’ll help your team take it from there.
Start with a readiness and scoping assessment: which Claude plan, who needs Claude Code, which connectors are allowed, and one workflow to pilot. We write down the scope, the access rules and the acceptance tests before anyone is invited.
A useful first workflowConnect Claude to Salesforce, SAP, NetSuite, HubSpot and your own applications. We build custom MCP servers and API integrations around the access and actions you need.
Your systems, connectedWe configure the settings your admins will own: SSO, provisioning, connector approvals and, on Enterprise, roles, retention and the Compliance API. Custom connectors run where your IT team decides and accept traffic from Anthropic’s published IP ranges, or run locally for Claude Desktop and Claude Code users.
A path into productionSeparate tracks for Claude admins, business users and engineers. Engineers set up Claude Code, design MCP tools, configure OAuth and test failure cases in a sandbox. This is independent technical training, not Anthropic certification.
A team that can own itCONNECTORS AND YOUR NETWORK
Where an MCP server runs decides what it can reach. Claude’s remote connectors are called from Anthropic’s cloud, not from the user’s laptop, so plan the network path before you build.
A custom connector in claude.ai, Claude Desktop, Cowork or the mobile apps reaches your MCP server from Anthropic’s infrastructure. The server must accept traffic from Anthropic’s published IP ranges, so a server on a private network needs a firewall allowlist.
On Team and Enterprise, only Owners can add a custom connector. Each member then connects with their own account, so Claude can reach only what that member can reach in the source system.
Local MCP servers in Claude Desktop’s configuration file, and stdio servers in Claude Code, run on the user’s computer and use its network. They suit developer tools and systems a laptop can already reach. Desktop-configured local servers are not available in claude.ai or Cowork.
Claude’s research feature can call connector tools without asking again, and Anthropic advises disabling write tools there. We ship read tools first and put an approval step on anything that changes a record.
ADMIN, SECURITY AND AUDIT
Team covers the basics. Enterprise adds the controls most security reviews ask for. We configure them with your identity and security teams, then test them with a restricted account.
SSO and JIT provisioning are available on Team and Enterprise. SCIM provisioning is an Enterprise feature, and so are role-based permissions, which control the features and connectors each group can use.
Enterprise Owners can export 180 days of audit logs. The Compliance API makes activity and conversation content, including Cowork and Claude Code sessions, available to your security tools. Only the Primary Owner can turn it on.
Enterprise supports custom retention periods of 30 days or more, customer-managed encryption keys and US-only inference. Anthropic states that prompts and results are not used to train its models by default.
Enterprise can limit access to approved IP ranges. Tenant Restrictions let your proxy block personal Claude accounts on the corporate network.
ENGINEERING ROLLOUT
Claude Code reads a codebase, edits files and runs commands. Roll it out like any developer tool with write access: identity, policy, shared configuration and a guided start.
Developers can sign in with a Team or Enterprise seat, a Claude Console account, or a cloud provider such as Amazon Bedrock, Google Cloud or Microsoft Foundry. Corporate proxies and LLM gateways are supported.
Managed settings let security teams fix permissions that local configuration cannot override. Managed MCP configuration can deploy a fixed server list or allow and deny specific servers.
A CLAUDE.md file describes the codebase and its conventions. A checked-in .mcp.json gives every developer the same approved MCP servers.
Anthropic suggests starting with questions about the codebase and small fixes, and asking Claude Code for a plan first. Scope widens as developers learn to review its work.
YOUR SYSTEMS. ONE USEFUL CONVERSATION.
Anthropic’s Connectors Directory covers many common apps. Your ERP, CRM and in-house systems usually need more: your fields, your approval rules, your network. We build custom MCP connectors that an Owner adds once and each member signs in to.
A connection that lets AI use your tools. You choose what it can read, which actions it can request, and when a person needs to approve.
Native connectors where they fit. Custom APIs and MCPs where you need more.
CAPABILITY, WITH ACCOUNTABILITY
We build the controls into the workflow, then make them understandable to the people responsible for it.
Define who can connect, which records they can reach, and the permissions each tool needs.
Separate reading from taking action. Add approval steps to the changes your team needs to review.
Plan logs, monitoring, evaluation and support so your team can investigate issues and improve the workflow.
Hosting, data processing and retention depend on the chosen platform and configuration. We document these with your team before launch.
A CLEAR PATH FORWARD
Claude set up for your organization, connected to the systems your people use, with a named owner who can run it in month six.
Choose a valuable workflow, understand your systems, and agree what success looks like.
Configure the platform, connect the right data, and test with the people who will use it.
Deploy the solution, run technical workshops, and put documentation and support in place.
Standalone technical workshops or hands-on training as part of your rollout.
HANDS-ON ENABLEMENT
Developers, technical leads and workspace administrators
Plan a Claude workspace rollout or build a connected workflow using approved APIs. Keep assistant configuration, MCP hosting and application permissions as separate decisions.
Choose a useful workflow and write acceptance criteria for its answers.
Design a read-only MCP tool with clear inputs, useful results and permission checks.
Test missing records, unauthorized requests and untrusted instructions in tool responses.
Document deployment, connector access and the steps for removing the connection.
A sandbox exercise, tool contract and operational handover checklist. Exercises use an agreed sandbox. Platform licenses and production approval are separate.
THE TEAM BEHIND YOUR ROLLOUT
AIONDATA builds and operates its own software products. We bring that engineering perspective to your implementation, integration and training.
Meet your delivery teamCalifornia leadership and an engineering team in India. Working sessions and handover planned around your time zone.
Our delivery process brings clear scope, documented decisions and a practical handover.
Source code, documentation and ownership terms agreed in your project scope. Vendor licenses remain separate.
BEFORE WE BEGIN
A little clarity before the first conversation.
No. AIONDATA is an independent consultancy and is not a member of Anthropic’s Claude Partner Network. We work from Anthropic’s published documentation and your own organization settings. Claude subscriptions and API usage are billed separately from our services.
Team suits groups of 2 to 150 that need SSO, central billing and connector controls. Enterprise adds SCIM, role-based permissions, audit logs, the Compliance API, custom retention and IP allowlisting, and bills usage at API rates on top of the seat fee. Your security team’s requirements usually decide it.
Yes. Anthropic bills for seats and usage, and a Claude subscription does not include Claude API access, which is billed through the Claude Console or your cloud provider. Our assessment, configuration, connector, training and handover work is scoped separately in writing.
Claude Enterprise offers US-only inference, and the Claude API can pin inference to the US per request. Through Amazon Bedrock, Google Cloud or Microsoft Foundry, regional options follow that platform. We document the full data path, including connectors, for your security review.
Yes, through MCP connectors. Check first whether the vendor offers an MCP server or connector for your edition. When you need your own fields, business rules or approval steps, we build a custom MCP server and add it as an organization connector.
Remote connectors are called from Anthropic’s cloud, so the server must accept Anthropic’s published IP ranges. If that is not acceptable, we can host an MCP layer in your cloud or DMZ that exposes only approved tools, or use local servers for Claude Desktop and Claude Code users. We agree the path with your network team first.
One workflow with a named owner, a clear input and output, and a person who reviews the result. Good examples are an account brief drawn from approved CRM fields, or order-status answers from your ERP. We keep write actions out of the first pilot unless there is an approval step.
Cowork is for knowledge work: you give Claude a goal and it works across files and connected tools. Claude Code is for software work in a repository. Team and Enterprise seats include both.
Yes. We run independent technical workshops for admins, business users and engineers, using a sandbox and your own examples. They do not lead to an Anthropic certification.
Your team does. We hand over connector code, runbooks, the access matrix, test cases and an admin checklist, and we train the named owner. Ongoing support can be added to the scope if you want it.
LET’S MAKE IT USEFUL
Tell us where you are with Claude. We’ll help you work out the right starting point.
Part of NVIDIA’s program for startups building with AI and accelerated computing.
About the program
Pravin’s experience includes Google and SmartBear. Sravan previously held leadership roles at JPMorgan Chase and First Republic Bank.
Meet the founders9110 Alcosta Blvd Ste H345, San Ramon, CA 94583
US-led delivery, with engineering in India. Supporting US and international organizations.No lengthy brief needed. We’ll work through it together.
AIONDATA provides independent implementation and consulting services. Product names belong to their respective owners.
